Dissertation > Excellent graduate degree dissertation topics show

Robot Localization in Fusion with Vision and Inertial Navigation

Author: LuDanZuo
Tutor: LiuJiLin
School: Zhejiang University
Course: Information and Communication Engineering
Keywords: Inertial navigation system Visual odometry Cumulative error Extended Kalman filter Simultaneous Localization and build terrain
CLC: TP242
Type: Master's thesis
Year: 2012
Downloads: 116
Quote: 0
Read: Download Dissertation

Abstract


Accurate mobile robot localization is an important field of academic research today, autonomous navigation. The limitations of traditional high accuracy positioning method based on the Global Positioning System and other external conditions of use. Use of robot vision system can achieve precise autonomous positioning applicable scenes, but are vulnerable to the impact of cumulative error, lack of integration of the inertial navigation system can effectively compensate for visual positioning method. This paper studies the fusion of INS robot visual the autonomous positioning method. Passed to dead reckoning attitude solution error due to traditional inertial navigation and positioning, after integral calculation error is cumulative affect the positioning accuracy and occurred \The reckoning avoid dead reckoning integration line acceleration using inertial navigation solution gesture wheeled odometer output real-time projection navigation coordinate system, to improve the positioning accuracy and stability of the inertial navigation. On the other hand, the existing visual mileage calculations estimated 3 heading, pitch and roll attitude angle is coupled angle estimation error in one direction will be projected onto the other two direction angle estimation, a long time cumulative , movement pose estimation serious direction deviation. In this paper, real-time extension Kalman filter pose estimation model using inertial navigation combined with the acceleration of gravity direction as a supplement, pose estimation is decoupled visual mileage of three directions, amendments pose estimation cumulative error; according to the state of motion using the fuzzy logic to adjust filtering parameters, adaptive filtering is estimated to reduce the impact of the acceleration noise, effectively improve the positioning accuracy and robustness of the visual odometry. Finally, inertial navigation and visual odometry is a local positioning methods, the lack of global information so that the accumulated error can not be corrected. In this paper, based on the inertial navigation visual simultaneous localization and terrain building proposed extended Kalman filter-based SLAM approach, local motion prediction, the use of INS data as matching SIFT features as a signpost data associated to amend the motion prediction, at the same time create a scene signpost map in an unknown environment to create and maintain the scene characteristics to further reduce the uncertainty of the position estimate, to make up for the lack of local positioning methods. The analysis and design of experiments to consider a variety of scenarios, total station surveying robot coordinates as the actual value, provides an effective means of error and positioning accuracy analysis. The experimental analysis shows that the fusion of inertial navigation and visual methods to improve the positioning accuracy, and shows the practical prospects of the proposed algorithm.

Related Dissertations

  1. Satellite Attitude Determination Based on Gyro and Star Sensor,V448.2
  2. Research on Key Techniques of Indoor Wireless Sensor Network Tracking,TN929.5
  3. For small unmanned helicopter SINS / GPS Integrated Navigation System,TN967.2
  4. Indoor mobile robot autonomous navigation and real-time map building,TP242
  5. Based on aircraft ejection seat posture Simulation Algorithm Research and Implementation,V244.212
  6. Formation of collision between satellites and anti-collision technology,V448.2
  7. Research on Attitude Determination and Magnetic Control for Small Satellites,V448.2
  8. Mobile robot based on GPS / DR integrated positioning technology research,TN967.2
  9. The Application Research on Estimation of Lithium-ion Battery SOC,TM912
  10. Research on Performance of Sensorless PMSM Vector Control System Based on EKF,TM341
  11. Anti-jamming Approach of Anti-ship Missile Using Inertial Navigation System Information,TN974
  12. Technology Research and Experiment of MEMS-IMU/Satellite Integrated Navigation,P228.4
  13. The Application of Fuzzy Adaptive Information Fusion to Integrated Vehicle Navigation System,U463.6
  14. Researh of Sins Virtual Technology in High Dynamic Environment,V249.322
  15. Research on Accelerometer Temperature Compensation System Based on DSP,TH824.4
  16. The Research of the Trajectory Prediction of Target Flight,TP391.41
  17. Research on Error Parameter Identification of Inertial Navigation Platform under the Condition of Vibration,V249.322
  18. Study on Induction Motor Vector Control System Based on State Parameter Estimation,TM346
  19. Investigation into INS Technique in Frequency Domain,V249.322
  20. Micromachined gyroscope random drift modeling,V241.5

CLC: > Industrial Technology > Automation technology,computer technology > Automation technology and equipment > Robotics > Robot
© 2012 www.DissertationTopic.Net  Mobile